Exploring clustering in SOMA

Abstract

During the developing phase of the new evolutionary algorithm (EA) or the analysis, several techniques or measurements are used to capture the inner dynamic of an algorithm. Besides the usual ones, for example, convergence graphs, population diversity, or complex networks, the scientists may also use clustering. Clustering analysis may naturally be used to analyze the grouping of individuals in swarm-based algorithms. This paper examines the possibilities of the clustering analysis for the Self-Organizing Migrating Algorithm with CLustering-aided migration (SOMA-CL). The algorithm is described in detail, together with several cluster analyses which can be used to investigate the behavior of the algorithm.